Finding an appropriate price using thousands of data points

If i have a lot of data points describing the price of a used car. How would I find the market value of the car (assuming that the price points in the data set are the only determinant used, and the basis of determination will be the frequency [higher the frequency the better] price data point occurence for that particular car).

A count of absolute value recurrences will not work, as I want to bucket numbers that are similar (less than 5% difference of total price) into one category.

Topic metadata

Category Data Science

1
answered at

It is not clear from your question if you are trying to predict value for many used cars as a function of different vehicle characteristics (like age of car, model, mileage) or predicting price of a single car.

Assuming the former (prediction using variables), you can set it up as a regression problem, where price is dependent variable and (age, model, mileage) are independent variables. Google "Boston housing price prediction" which is a similar application in different domain.

If you have only prices data (ignoring vehicle characteristics), you are looking for summary statistics. You can start with the mean (average) or median price. You can then it little more complex by modelling the price as some reasonable statistical distribution (normal or Poisson) and then take typical summary statistics on the distribution as estimate (say inter-quantile range).

About

Geeks Mental is a community that publishes articles and tutorials about Web, Android, Data Science, new techniques and Linux security.